RegEx to remove lines from MySQL dump - regex

Question

CREATE TABLE `transaction` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`description` text NOT NULL,
PRIMARY KEY (`id`),
FULLTEXT KEY `description` (`description`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
How can I remove FULLTEXT line from MySQL dump above and comma on the line before so it looks something like this:
CREATE TABLE `transaction` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`description` text NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
The removal should be easy using sed but I'm not sure how to remove this comma on the line above so the dump is successfully imported:
sed -i '/FULLTEXT KEY.*/d' dump.sql
Sometimes there is also more columns with FULLTEXT index:
CREATE TABLE `entity` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`company_name` varchar(255) NOT NULL DEFAULT '',
`description` text NOT NULL,
PRIMARY KEY (`id`),
FULLTEXT KEY `company_name` (`company_name`),
FULLTEXT KEY `description` (`description`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;

Answer 1

Sed script:
#!/bin/bash
sed -e '/FULLTEXT/d' |
sed -ne '
/ENGINE=InnoDB/!{H}
/ENGINE=InnoDB/{x; s/,[ \t]*$//; p; }
${g;p;}
'
Input:
CREATE TABLE `transaction` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`description` text NOT NULL,
PRIMARY KEY (`id`),
FULLTEXT KEY `description` (`description`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE `entity` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`company_name` varchar(255) NOT NULL DEFAULT '',
`description` text NOT NULL,
PRIMARY KEY (`id`),
FULLTEXT KEY `company_name` (`company_name`),
FULLTEXT KEY `description` (`description`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
Sample Run:
/home/user> ./1.sed < input
CREATE TABLE `transaction` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`description` text NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE `entity` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`company_name` varchar(255) NOT NULL DEFAULT '',
`description` text NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;

Answer 2

sed is for simple substitutions on individual lines, that is all. For anything else you should use awk for clarity, brevity, portability, efficiency, robustness and most other desirable qualities of software. All of the sed constructs to do anything other than s, g, and p (with -n) became obsolete in the mid-1970s when awk was invented and exist today just for mental exercise.
Given this input:
$ cat file
CREATE TABLE `transaction` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`description` text NOT NULL,
PRIMARY KEY (`id`),
FULLTEXT KEY `description` (`description`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE `entity` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`company_name` varchar(255) NOT NULL DEFAULT '',
`description` text NOT NULL,
PRIMARY KEY (`id`),
FULLTEXT KEY `company_name` (`company_name`),
FULLTEXT KEY `description` (`description`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
With GNU awk for multi-char RS:
$ awk -v RS=',\\s*FULLTEXT[^\n]*)' -v ORS= '1' file
CREATE TABLE `transaction` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`description` text NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE `entity` (
`id` mediumint(9) NOT NULL AUTO_INCREMENT,
`company_name` varchar(255) NOT NULL DEFAULT '',
`description` text NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
Since you are using GNU sed (for -i) I assume you will have no problem using GNU awk and if you need inplace editing add -i inplace at the start.

Answer 3

awk -vRS=";" 'NF{gsub(/),\n +FULLTEXT.*)/,")\n)",$0);$0=$0";"}1' file